5 min read EU AI Act
With the increasing integration of artificial intelligence (AI) across different sectors, it is essential for us to grasp and adjust to the changing regulatory environment. The AI Act implemented by the European Union establishes varying responsibilities for AI suppliers and consumers depending on the risk level associated with AI technologies. We will examine the distinct risk categories specified in the AI Act, examine concrete instances, and discuss strategies for your organization to successfully conform to these fresh regulations.
Unacceptable Risk: Systems That Will Be Banned
The AI Act categorizes certain AI systems as posing an "unacceptable risk." These are systems that are considered threats to individuals and society and will be outright banned. Examples include:
- Cognitive Behavioural Manipulation: AI systems that manipulate people's thoughts and behaviours, especially those targeting vulnerable groups, such as voice-activated toys that encourage dangerous activities in children.
- Social Scoring: Systems that classify people based on their behaviour, socio-economic status, or personal characteristics.
- Biometric Identification and Categorisation: Real-time and remote biometric identification systems, such as facial recognition technology, which raise significant privacy and ethical concerns.
There are some exceptions for law enforcement purposes. Real-time remote biometric identification systems may be permitted in a limited number of serious cases, while post-remote biometric identification systems (where identification occurs after a delay) will be allowed to prosecute serious crimes, but only with court approval.
High Risk: Systems Requiring Rigorous Oversight
AI systems that negatively affect safety or fundamental rights are classified as high-risk. These systems are divided into two main categories:
- Product-related AI Systems: These include AI systems integrated into products regulated under the EU’s product safety legislation, such as toys, aviation, automobiles, medical devices, and lifts.
- Sector-specific AI Systems: These systems operate in areas that require stringent oversight and must be registered in an EU database. Key areas include:
- Management and operation of critical infrastructure
- Education and vocational training
- Employment, worker management, and access to self-employment
- Access to essential private and public services
- Law enforcement
- Migration, asylum, and border control management
- Assistance in legal interpretation and application of the law
Before high-risk AI systems are introduced to the market and during their operation, they will go through evaluations. Furthermore, people will be allowed to report concerns about high-risk AI systems to specific national bodies.
Transparency Requirements: Ensuring Accountability
Generative AI models, such as ChatGPT, won't fall under the high-risk category but must adhere to strict transparency requirements and EU copyright laws.
Key obligations include:
- Disclosure: Clearly indicating that content was generated by AI.
- Content Control: Designing models to prevent the generation of illegal content.
- Copyright Compliance: Publishing summaries of copyrighted data used in training the AI.
High-impact general-purpose AI models, like GPT-4, which might pose systemic risks, must undergo thorough evaluations, and any serious incidents must be reported to the European Commission.
Adapting to the AI Act: Strategies for Your Company
To navigate the new regulatory landscape effectively, here's what we recommend for your company:
- Risk Assessment: Conduct comprehensive risk assessments of all AI systems to determine their classification under the AI Act. This will help you understand the specific obligations and compliance requirements.
- Compliance Infrastructure: Develop robust compliance frameworks and processes to ensure adherence to the AI Act’s regulations. This includes setting up mechanisms for continuous monitoring and reporting.
- Transparency Measures: Implement transparent practices, particularly for generative AI models. Ensure that all AI-generated content is appropriately labeled and that data sources are disclosed in compliance with copyright laws.
- Stakeholder Engagement: Engage with stakeholders, including customers, employees, and regulatory bodies, to foster trust and demonstrate commitment to ethical AI practices.
- Training and Education: Invest in training programs for employees to stay updated on the latest regulatory requirements and best practices in AI governance.
By understanding and adapting to the AI Act, we can ensure compliance while fostering innovation and trust in our AI systems. Embracing these regulations will pave the way for safer and more ethical AI deployment, benefiting both our businesses and society at large.
Get in touch with us:
AI and Emerging Technology- Strategy, compliance and readiness:
https://gottdata.com/services/ai-and-tech
Follow us on LinkedIn: Gott Data